Latest AI News

Elon Musk Testifies xAI Used OpenAI Models in Training Grok

In a California federal courtroom on Thursday, Elon Musk told a judge that his AI startup xAI employed OpenAI’s models to develop its own system, Grok, through a practice known as model distillation. Musk said the technique is common across the industry, answering “partially” when asked if xAI directly distilled OpenAI technology. The testimony highlights a growing debate over the legality and ethics of AI model sharing, with companies like OpenAI, Anthropic and Google warning of potential intellectual‑property violations.

OpenAI Introduces Advanced Account Security for ChatGPT and Codex Users

OpenAI announced Thursday that it will roll out an optional Advanced Account Security tier for ChatGPT and Codex accounts. The new feature replaces passwords with physical security keys or passkeys, eliminates email and SMS recovery, and forces tighter sign‑in windows. Aimed at high‑risk users such as journalists, elected officials and researchers, the measure also blocks support staff from accessing recovery options, limiting attackers’ social‑engineering avenues. OpenAI partnered with Yubico to offer discounted YubiKey bundles and will require members of its Trusted Access for Cyber program to enable the tier by June 1.

Google to Upgrade Cars with Built‑In Google to Gemini AI Assistant

Google announced that its Gemini AI chatbot will be rolled out to vehicles equipped with Google built‑in, replacing the current Google Assistant. The upgrade will arrive via a software update, reaching both new and existing models, starting with English‑language cars in the United States. Gemini promises more natural conversation, real‑time trip updates, and deeper integration with apps such as Gmail and Calendar, while also offering electric‑vehicle data like battery level. General Motors has already confirmed Gemini will appear in millions of its 2022‑plus models, but Google says the rollout will not be limited to GM vehicles.

OpenAI explains lingering goblin references in its AI models

OpenAI has detailed why its language models occasionally mention goblins, gremlins and other mythic creatures. The issue first surfaced with the GPT-5.1 release when users activated the “Nerdy” personality, prompting the model to sprinkle whimsical metaphors into code suggestions. Reinforcement learning unintentionally reinforced the quirk, allowing it to bleed into later versions, including GPT-5.5’s Codex tool, despite the company’s effort to suppress the behavior. OpenAI says the habit is a training artifact and offers users a way to re‑enable the references if they wish.

Microsoft, OpenAI Revise Partnership, Allowing Multi‑Cloud AI Model Access

Microsoft announced on Monday that its long‑standing deal with OpenAI has been rewritten to let the AI startup sell its models on any cloud provider, including Amazon Web Services. The amendment extends Microsoft’s revenue‑share rights through 2032, removes an artificial‑general‑intelligence clause, and keeps Azure as OpenAI’s primary partner while opening the door for rivals to host the latest models. The shift follows months of tension over exclusivity and revenue terms, and it reshapes the financial and strategic stakes for both companies.

OpenAI moves ChatGPT model selector into prompt bar, adds hidden Pro option

OpenAI has quietly revamped the ChatGPT interface for Plus, Pro and Business users, placing the model selector—Instant, Thinking and Configure—directly beside the prompt box. The change makes it easier for subscribers to pick the right version of GPT for their tasks, with Instant delivering speed, Thinking offering deeper reasoning, and a hidden Pro tier unlocking the most powerful GPT‑5.5 engine. A new auto‑select switch and an extended thinking mode also appear in the Configure menu, giving users finer control over performance and resource usage.